Chinese J. Chem. Eng.

, 14(6) 770779 (2006)

Calculation of Viscosities of Liquid Mixtures Using Eyrings Theory

in Combination with Cubic Equations of State*

JI Weirong()a,** and D. A. Lempeb

a
College of Chemical Engineering and Materials Science, Zhejiang University of Technology, Hangzhou 310014,
China
b
Department of Engineering Science, Martin-Luther University of Halle-Wittenberg, D-06099, Halle, Germany

Abstract Cubic equations of state (EOS) have been combined with the absolute rate theory of Eyring to calculate
viscosities of liquid mixtures. A modified Huron-Vidal gE-mixing rule is employed in the calculation and in com-
parison with the van Laar and the Redlich-Kister-type mixing rule. The EOS method gives an accurate correlation
of liquid viscosities with an overall average deviation less than 1% for 67 binary systems including aqueous solu-
tions. It is also successful in extrapolating viscosity data over a certain temperature range using parameters obtained
from the isotherm at a given temperature and in predicting viscosities of ternary solutions from binary parameters
for either polar or associated systems.
Keywords viscosity, equation of state, mixing rule, absolute rate theory

1 INTRODUCTON model directly with activity coefficient equations. Wei

The viscosity, particularly that of liquid mixtures,
is very important in engineering calculations involved
in the process design for petroleum and other chemical
industries. Since the successful development of a
one-parameter equation for correlating the liquid vis-
cosity of nonpolar mixtures by Grunberg and Nissan[1],
many other models have been proposed. Most of them
are based on the corresponding state principle, the
absolute rate theory of Eyring[2], or the free volume
theory. A detailed review about viscosity calculations
was given by Poling et al.[3]. Among the correspond-
ing state models, the generalized corresponding state
principle (GCSP) proposed by Teja and Rice[4] is im-
portant. By using two reference fluids and in combi-
nation with a van Laar-type of mixing rule, it can be
used to correlate viscosities of nonaqueous polar bi-
nary liquid mixtures over a certain temperature range.
Unfortunately, it is not adequate for strongly associ-
ated systems such as aqueous solutions, where a
maximum exists in the viscosity-composition curve[5]
and its extension to ternary systems is not very suc-
cessful either[6]. For the methods based on the Ey-
rings theory, the estimation of the activation Gibbs
energy of flow plays an important role. McAllister [7]
calculated this quantity by using an empirical cubic
composition-dependent function. The resultant model
is very successful in correlating viscosities of liquid
mixtures. However, its parameters are strongly tem-
perature-dependent and for ternary systems, additional
ternary parameters are necessary. Another application
of Eyrings theory is the combination of the viscosity
and Rowley[8] developed a local composition model.
They used the nonrandom two liquids (NRTL) equa-
tion to estimate the entropy part of the excess activa-
tion Gibbs energy of flow. However, the viscosity
calculation needs not only the NRTL parameters ob-
tained from vapor-liquid equilibrium (VLE) data but
also the experimental excess enthalpies and the liquid
densities. Wu[9] and Chevalier et al.[10] estimated the
excess activation Gibbs energy of flow by using a
group-contribution (GC) concept such as the universal
quasi-chemical functional group activity coefficient
(UNIFAC) method. The proposed UNIFAC-VISCO
model can give a satisfactory prediction for nonpolar
systems. Unfortunately, the predictions are not always
reliable for polar mixtures, particularly for aqueous
solutions. A similar strategy was used by Chao et
al.[11,12] to calculate viscosities and vapor-liquid equi-
libria simultaneously, but only nonaqueous mixtures
were investigated. Another way of estimating the ac-
tivation Gibbs energy of flow is by using the equa-
tions of state. Lee et al.[13] combined the Eyrings the-
ory with Patel-Teja equation of state (PTEOS) to form
an Eyring-Patel-Teja viscosity model. It can correlate
the viscosities of binary systems very well even at
elevated pressures and is successfully extended to ter-
nary systems, however, the results for aqueous solu-
tions are not very satisfactory possibly because the
Redlich-Kister-type (RK) mixing rule was em-
ployed[14], which is not very suitable for multicompo-
nent systems[15]. A successful model based on the free
volume theory was proposed by Liu et al.[5]. With two

Received 2005-09-28, accepted 2006-04-15.

* Supported by the Deutsche Forschungsgemeinschaft (LE 886/4-1) and the Foundation of Zhejiang Province for Scholars
Returned from Abroad.
** To whom correspondence should be addressed. E-mail: weirong.ji@zjut.edu.cn
 Calculation of Viscosities of Liquid Mixtures Using Eyrings Theory in Combination with Cubic Equations of State 771

temperature-independent parameters, the free volume (Table 1) calculated using the experimental viscosity
model can describe the viscosities of liquid mixtures data in comparison with the experimental excess en-
satisfactorily over a certain temperature range even for thalpies (hE)[31] and the excess Gibbs energies obtained
aqueous systems. from the experimental VLE data, respectively. It can be
As discussed above, a reasonable model for the seen that a rough linear relationship between the excess
calculation of the liquid viscosity is able to describe its activation Gibbs energy of flow and the excess Gibbs
temperature dependence, can be extended to multi- free energy or the excess enthalpy exists for nonaque-
component systems, and is suitable for strongly associ- ous systems, which provides to some extent a back-
ated solutions. In this study, the cubic equations of state ground for relating the transport properties to the ther-
in combination with a modified Huron-Vidal gE-mixing modynamic properties (quasi-thermodynamic consid-
rule (see Appendix) are incorporated into the Eyrings eration), and also provides a simple way to incorporate
theory to calculate liquid viscosities of either nonpolar the flexible structure of thermodynamic models into
or polar systems including aqueous solutions. The tem- the absolute rate theory of Eyring. Nevertheless, a di-
perature extrapolation and the prediction of viscosities rect prediction of viscosities using VLE or hE data
for multicomponent systems are investigated. The van seems to be impossible, particularly for aqueous sys-
Laar-type (VL) mixing rule[16] and the RK mixing tems. The viscosity equation for liquid mixtures is
rule[14] are used in the calculation for comparison. obtained by substituting Eq.(6) into Eq.(4)

v = ( v)id exp g
E
2 VISCOSITY MODEL (7)
According to the absolute rate theory of Eyring, RT
the dynamic viscosity can be estimated by
g #
v = N A h0 exp (1)
RT
where, g# is the activation Gibbs energy of flow that
is required to remove molecules within the fluid from
their energetically most favorable state to the activated
state and the quantities v, NA and h0 represent the mo-
lar volume, the Avogadro and the Planck constant,
respectively. Eq.(1) can be used either for pure fluids
or for mixtures. For a liquid solution the activation
Gibbs energy of flow can be separated into a hypo-
thetical ideal solution part and an excess part.
# # id #E
g = g + g (2)
From the ideal activation Gibbs energy of flow the
viscosity of an ideal solution is given
g # id
( v )
id
= N A h0 exp (3)
RT
The combination of Eqs.(2), (3), and (1) follows the
viscosity equation of liquid mixtures
#E
v = ( v)id exp g (4)
RT
where (v)id can be estimated by a simple combination
of the (kinematic) viscosities of the pure fluids [(v)oi]
ln( v)id =
xk ln( v)ok
k
(5)

There are several ways to calculate the excess ac-

tivation Gibbs energy of flow (g#E) in the literature.
One such calculation is to relate it to the excess Gibbs
free energy (gE). Wei and Rowley[8] proposed a simple
relation
Figure 1 Comparison of the excess activation Gibbs energy
g # E = g E (6) of flow with the experimental excess enthalpy and the excess
where is a proportional factor. Figs.1(a) and 1(b) Gibbs energy obtained from the experimental VLE data
nonaqueous; aqueous; fitting nonaqueous
show the g#E values at xi=0.5 for 20 binary systems
Chinese J. Ch. E. 14(6) 770 (2006)
772 Chinese J. Ch. E. (Vol. 14, No.6)

Table 1 Correlation of viscosity isotherms of binary systems

, oC3EOS PREOS PT EOS
System
rel ( ) rel ( )VL rel ( )2p rel ( )3p rel ( ) rel ( )VL rel ( )2p rel ( )3p rel ( )RK
[17]
acetone+cyclohexane 25 1.39 0.57 0.65 0.34 5.08 0.62 0.64 0.30 0.75
acetone+ethanol[17] 25 2.28 0.45 0.27 0.17 10.35 0.50 0.29 0.17 0.57
acetone+hexane[17] 25 0.76 0.71 0.74 0.34 5.78 0.61 0.93 0.30 0.72
acetone+methanol[17] 25 1.69 0.20 0.23 0.23 14.14 0.21 0.24 0.23 0.25
acetone+propan-2-ol[17] 25 3.28 0.79 6.55 0.19 10.94 0.81 6.83 0.28 1.20
acetone+carbon tetrachloride[17] 25 1.47 0.91 0.92 0.91 4.98 1.02 0.92 0.92 0.95
acetone+water[18] 25 2.07 1.29 11.96 5.39 11.22 1.30 12.39 5.74 1.87
benzene+cyclohexane[19] 25 2.30 0.30 0.28 0.20 2.94 0.31 0.28 0.33 0.38
benzene+decane[19] 25 3.13 0.62 0.98 0.46 4.32 0.72 0.87 0.98 0.67
benzene+ethanol[20] 25 0.89 0.21 0.21 0.21 2.43 0.18 0.22 0.18 0.40
benzene+hexane[21] 25 0.61 0.20 0.18 0.17 1.25 0.21 0.18 0.15 0.09
benzene+toluene[21] 25 0.98 0.11 0.13 0.11 1.14 0.11 0.14 0.11 0.08
butan-1-ol+ethyl acetate[22] 25 3.36 0.35 2.05 0.31 4.39 0.36 2.33 0.35 0.98
butan-1-ol+ethyl benzene[22] 25 3.85 0.57 0.70 0.47 2.15 0.58 2.74 0.49 0.30
butan-1-ol+xylene[22] 25 4.00 0.27 0.45 0.21 3.55 0.28 1.54 0.20 0.13
chlorobenzene+benzyl alcohol[23] 30 2.98 0.66 0.58 0.57 0.83 0.64 0.58 0.64 0.41
chlorobenzene+1-hexanol[24] 30 3.69 0.28 0.30 0.28 1.71 0.28 0.31 0.29 0.29
chloroform+benzene[24] 25 0.84 0.35 0.46 0.36 4.75 0.37 0.46 0.35 0.28
chloroform+methanol[17] 25 2.31 1.33 3.56 1.82 4.36 1.60 3.86 1.92 0.84
chloroform+toluene[24] 25 0.77 0.13 0.21 0.13 3.31 0.13 0.26 0.15 0.13
cyclohexane+propan-2-ol[17] 25 1.23 0.53 0.49 0.46 1.60 0.59 0.49 0.57 0.41
dimethylsulfoxide+chloroform[17] 25 2.15 3.10 3.28 3.17 10.09 3.11 3.24 3.17 2.87
dimethylsulfoxide+methanol[17] 25 3.71 0.46 0.38 0.47 19.58 0.47 0.48 0.47 0.48
ethanol+cyclohexane[17] 25 0.24 0.74 0.61 0.55 2.081 0.47 0.60 0.61 1.73
ethanol+dimethylformamide[17] 25 2.49 0.16 0.22 0.22 19.08 0.29 1.35 0.39 2.40
ethanol+heptane[25] 25 0.26 1.05 1.35 1.13 3.22 1.83 1.37 1.34 1.12
ethanol+octane[26] 25 2.22 0.60 1.18 0.55 4.67 0.62 1.05 1.13 0.66
ethanol+pentane[26] 25 2.40 2.80 1.86 1.46 2.01 2.25 1.83 1.81 1.31
ethanol+propan-1-ol[18] 25 1.31 0.35 0.33 0.12 3.67 0.25 0.30 0.13 0.18
ethanol+propan-2-ol[17] 25 0.22 0.24 0.24 0.24 4.66 0.24 0.24 0.24 0.24
ethanol+carbon tetrachloride[27] 25 0.84 0.57 1.65 0.47 2.31 0.83 1.78 0.63 1.16
ethanol+water[18] 30 2.37 2.21 1.64 1.62 11.71 2.31 1.64 1.62 4.20
ethan-1,2-diol+water[18] 25 2.94 0.74 0.52 0.51 11.18 0.79 0.51 0.52 0.98
heptane+benzene[20] 25 1.03 0.45 0.44 0.38 0.78 0.45 0.44 0.41 0.46
hexane+benzyl alcohol[25] 30 2.58 2.87 2.89 2.87 1.50 2.95 2.90 2.90 3.12
hexane+chlorobenzene[28] 30 0.69 0.56 0.71 0.74 0.36 0.51 0.71 0.74 0.76
hexane+chloroform[17] 25 1.07 0.33 0.35 0.32 3.62 0.53 0.35 0.31 0.31
hexane+cyclohexane[17] 25 0.85 0.30 0.29 0.28 1.62 0.30 0.29 0.29 0.28
hexane+decane[18] 25 2.00 0.60 0.59 0.60 3.66 0.60 0.60 0.63 0.52
hexane+ethanol[17] 25 0.53 0.82 0.74 0.23 2.75 0.34 0.77 0.21 1.04
hexane+heptane[18] 25 0.40 0.14 0.14 0.14 1.30 0.13 0.14 0.14 0.06
hexane+hexan-1-ol[28] 30 2.58 1.10 1.12 1.05 1.02 1.09 1.18 1.11 0.96
hexane+nonane[18] 25 1.25 0.36 0.40 0.36 2.59 0.36 0.39 0.38 0.29
hexane+octane[18] 25 1.86 0.43 0.42 0.42 1.52 0.43 0.43 0.43 0.35
hexane+toluene[28] 30 0.22 0.79 0.83 0.87 0.31 0.99 0.82 0.80 0.81
hexan-1-ol+benzyl alcohol[23] 30 4.39 0.59 0.45 0.47 0.27 0.53 0.45 0.47 0.37
methanol+ethanol[17] 25 0.62 0.22 0.31 0.19 11.16 0.21 0.32 0.17 0.23
methanol+heptane[26] 25 1.00 0.56 0.64 0.49 6.88 0.57 0.66 0.51 0.58
methanol+hexane[26] 25 2.25 0.77 1.32 0.77 5.51 0.77 1.33 0.78 0.95
methanol+pentane[26] 25 3.95 1.08 1.06 0.42 5.27 0.99 1.50 1.58 1.18
methanol+propan-1-ol[18] 25 1.17 1.28 1.29 1.28 6.84 1.28 1.28 1.27 1.26
methanol+propan-2-ol[17] 25 0.62 0.27 0.27 0.27 9.50 0.27 0.27 0.27 0.27
methanol+toluene[29] 25 0.58 0.89 1.69 1.04 5.45 0.91 1.76 1.08 0.77

December, 2006
 Calculation of Viscosities of Liquid Mixtures Using Eyrings Theory in Combination with Cubic Equations of State 773

Table 1 (Continued)
, oC3EOS PREOS PT EOS
System
rel ( ) rel ( )VL rel ( )2p rel ( )3p rel ( ) rel ( )VL rel ( )2p rel ( )3p rel ( )RK
[18]
methanol+water 30 1.79 1.13 0.59 0.62 13.50 1.19 0.58 0.64 1.62
methyl acetate+chloroform[17] 25 2.62 0.59 1.01 0.60 2.86 0.65 0.99 0.64 0.30
methyl acetate+hexane[17] 25 1.72 0.30 0.31 0.22 1.94 0.31 0.31 0.29 0.27
propan-1-ol+heptane[26] 25 1.91 0.86 0.64 0.59 1.80 0.82 0.64 0.60 0.43
propan-1-ol+hexane[26] 25 1.02 1.80 1.71 1.54 0.69 1.73 1.72 1.52 1.57
propan-1-ol+octane[26] 25 3.98 0.95 0.64 0.59 3.27 1.00 0.64 0.65 0.52
propan-1-ol+pentane[26] 25 1.23 2.84 2.61 2.52 1.23 2.72 2.62 2.43 1.81
propan-1-ol+water[26] 25 1.75 2.69 3.57 0.30 5.58 2.94 5.53 0.58 5.00
carbon tetrachloride+propan-1-ol[17] 25 0.84 0.42 0.46 0.41 1.39 0.39 0.46 0.37 0.29
carbon tetrachloride+benzene[30] 25 0.63 0.05 0.20 0.13 3.56 0.04 0.20 0.20 0.18
carbon tetrachloride 25 0.84 0.09 0.10 0.12 4.31 0.09 0.26 0.25 0.32
+cyclohexane[30]
triethylamine+chloroform [17] 25 0.84 2.21 2.22 2.04 6.80 2.24 2.36 2.07 2.11
triethylamine+methanol [17] 25 1.25 0.31 3.42 0.60 5.92 0.31 3.93 0.78 0.83
toluene+decane [19] 25 3.06 0.46 0.51 0.48 4.91 0.49 0.52 0.52 0.47
overall average deviation 1.76 0.78 1.18 0.71 4.85 0.81 1.31 0.78 0.89
System involved in Figs.1(a) and 1(b).
Note: rel ( x ) = 1 xcal xexp 100 % .

where the excess Gibbs free energy gE can be calcu-
lated by the following equation
g E = RT x ( ln
k
k
*
k ln o*k )
Here, i* and o*i represent the fugacity coefficients
of the component i in mixtures and at a pure liquid
state, respectively. They can be evaluated using the
equations of state. In this study, the optimized cubic
3-parameter equation of state (oC3EOS) proposed in
the literature[32], which can describe pvT-behavior sat-
isfactorily over a wide temperature and pressure range,
and the Peng-Robinson equation of state (PR EOS) in
combination with a modified Huron-Vidal gE-mixing
rule[33] based on the zero pressure reference state and
the van Laar-type mixing rule has been incorporated
into Eq.(7) to calculate liquid viscosities. The details
on the oC3EOS and the mixing rules used are given in
the Appendix. Thus, the viscosity model proposed
comprises Eqs.(5), (7) and (8).
3 CALCULATION
3.1 Correlation of viscosities for binary systems
Viscosities of 67 binary liquid systems including
strongly associated solutions are investigated, most of
them being isothermal data at 25 (with a few at
30). The procedure of viscosity calculations is as
follows. First, the kinematic viscosity (v)id of an ideal
solution is estimated using Eq.(5). Second, the excess
Gibbs free energy (gE) is evaluated using Eq.(8), in
which the fugacity coefficients are calculated using
EOS methods. The viscosity of liquid mixtures are
then obtained by substituting the value of (v)id and gE
into Eq.(7), in which the proportional factor is taken
as a constant equal to 0.25, which was suggested by
Wei and Rowley[8]. The molar volume of both pure
liquids and liquid mixtures used in Eqs.(5) and (7) are
(8)
predicted using the equations of state. In the
gE-mixing rule the NRTL equation is used to estimate
the excess Gibbs free energy of mixtures at zero pres-
sure. The parameters in the mixing rules are optimized
using the following objective function:
1 N exp cal
Fobj =
N
1 exp
(9)
n
In the calculation the nonrandomness parameter
in the gE-mixing rule is either set as constant value
of 0.2 (2p gE-mixing rule) or fitted together with the
NRTL energy parameters (Aij and Aji) (3p gE-mixing
rule) using the experimental viscosity data. The results
of the oC3EOS using the two variants of gE-mixing
rule and the van Laar-mixing rule are listed in Table 1
in comparison with those of the PREOS. During the
viscosity correlation, the densities of the liquid mix-
tures are estimated simultaneously using the correlated
parameters. The density prediction results of both the
EOS with the 3p gE-mixing rule are also presented.
As shown in Table 1, by the use of the 3p
gE-mixing rule, both the oC3EOS and the PREOS can
reproduce experimental liquid viscosity values very
well not only for nonpolar liquid mixtures but also for
polar liquid mixtures and even for strongly associated
systems. The overall average relative deviation
amounts to 0.7% for the oC3EOS and 0.8% for the
PREOS. For the majority of the systems, the use of a
constant nonrandom parameter (0.2) leads to no
obvious loss of accuracy. Only in the case of some

Chinese J. Ch. E. 14(6) 770 (2006)

774 Chinese J. Ch. E. (Vol. 14, No.6)

strongly polar systems, the adjustment of is neces- parameters obtained from the isotherm at 25 ,
sary. It is remarkable that the van Laar-type mixing whereas the dash-dot lines show the results calculated
rule is as flexible as the gE-mixing rule in the correla- with parameters from the isotherm at 10. It is re-
tion, and because of its composition asymmetry, it can markable that there is no considerable loss of the ex-
represent more accurately the experimental viscosities trapolation accuracy even if the temperature range is
for the system acetone+water, of which the composi- doubled. The average deviations of the 11 systems for
tion dependence of the viscosity is very strong and both the oC3EOS and PREOS are shown in Table 2 in
asymmetric. However, it should be noted that the van comparison with those of the PTEOS with the RK
Laar-mixing rule needs a very large binary interaction mixing rule. For nonaqueous systems, the three EOS
correction term for aqueous systems. The value at a give very similar results. For aqueous systems, the
certain composition range is larger than 1 (Table A1), difference of extrapolations obviously depends mainly
which is too large from the theoretical point of view on the flexibility of the mixing rule used.
and, as shown later, could lead to problems in the ex-
tension to multicomponent systems. The oC3EOS pre-
dicts saturated liquid densities more accurately than
the PREOS. The overall average deviation of the es-
timated liquid densities is 1.8% for the oC3EOS and
4.9% for the PREOS. However, no obvious advantage
of the oC3EOS over the PREOS is shown in the dy-
namic viscosity correlation. It seems that the accurate
description of the liquid molar volume plays no sig-
nificant role in this case. Possibly this resulted from
the fact that the molar volume appears on either sides
of Eq.(7) and therefore, its influence becomes rela-
tively small.
For comparison, similar calculations are carried
out using the PT EOS together with the RK mixing
rule proposed by Adachi and Sugie[14]. As shown in
Table 1, the asymmetric RK mixing rule gives results
Figure 2 Prediction of viscosities of the system hex-
very similar to those of the van Laar-mixing rule. ane+hexan-1-ol using the oC3EOS with the 3p and 2p
Only for aqueous systems, the deviation of dynamic variants of gE-mixing rule and the van Laar-mixing rule
viscosities is slightly larger. The overall deviation of 30; 40; 50; 60;
E E
this model amounts to 0.89%, only slightly larger than oC3EOS 3p g ; - - - oC3EOS 2p g ; ------ oC3EOS VL
those of the other two EOS together with the van
Laar-mixing rule.

3.2 Prediction of viscosities for binary systems

The liquid viscosity is usually strongly tempera-
ture dependent. It is of great importance if a model
can be used to extrapolate the viscosities over a cer-
tain temperature range using the information obtained
at a given temperature. The viscosities of 11 binary
systems with a total of 38 isotherms have been pre-
dicted using the parameters obtained from the correla-
tion of viscosity isotherms at 25 (for a few systems
at 30), as described above. The extrapolation is sat-
isfactory, particularly for the nonaqueous systems.
Fig.2 shows the results of the system hex-
ane+hexan-1-ol using the oC3EOS in combination
with the 3p gE-mixing rule (solid lines), the 2p Figure 3 Prediction of viscosities of the system metha-
gE-mixing rule (dashed lines) and the van Laar-mixing nol+water using the oC3EOS with the 3p and 2p variants of
rule (dotted lines). The predicted viscosities agree well gE-mixing rule and the van Laar-mixing rule
10; 20; 25; 40; 50;
with the experimental values for all the three mixing E E*
oC3EOS 3p g ; ---- oC3EOS 3p g ;
rules. Reasonable predictions are also obtained for the E ------
- - - - oC3EOS 2p g ; oC3EOS VL
aqueous systems. In Fig.3, the solid lines represent the
dynamic viscosity values of the system metha- 3.3 Prediction of viscosity for ternary systems
nol+water predicted using the 3p gE-mixing rule with The viscosities of ternary systems are predicted
December, 2006
 Calculation of Viscosities of Liquid Mixtures Using Eyrings Theory in Combination with Cubic Equations of State 775

Table 2 Prediction (extrapolation) of viscosities for binary systems

, oC3EOS PREOS PT EOS

Systems
rel ( )VL rel ( )2p rel ( )3p rel ( )VL rel ( )2p rel ( )3p rel ( )RK

hexane + toluene[28] 40 2.01 2.03 1.98 2.01 1.99 1.97 1.69

50 1.13 1.07 1.13 1.14 1.06 1.14 1.39
60 1.86 1.75 1.78 1.85 1.68 1.80 2.00
hexane + chlorobenzene[25] 40 1.94 1.86 1.76 1.64 1.84 1.77 1.86
50 1.09 0.88 0.92 0.66 0.90 0.91 0.74
60 0.98 0.90 0.92 0.82 0.92 0.91 0.96
hexane + benzyl alcohol[25] 40 2.75 3.15 3.15 2.77 3.02 3.14 3.41
50 1.76 2.77 2.76 2.19 2.63 2.69 2.85
60 3.75 4.60 4.61 4.08 4.49 4.53 4.67
hexane + hexan-1-ol[28] 40 1.90 2.36 2.33 1.73 2.38 2.34 1.89
50 1.41 1.18 1.17 1.58 1.19 1.19 1.58
60 3.77 2.57 2.60 4.11 2.54 2.57 3.62
chlorobenzene+benzyl 40 2.89 3.30 3.30 3.08 3.32 3.35 2.55
alcohol[23] 50 2.66 3.12 3.13 2.72 3.14 3.15 2.34
60 3.09 3.49 3.51 3.17 3.50 3.48 2.71
chlorobenzene+hexan-1-ol[23] 40 2.65 2.98 2.98 2.59 2.99 3.00 2.57
50 1.39 0.74 0.74 1.49 0.72 0.71 1.47
60 3.02 1.99 1.99 3.14 1.92 1.93 3.07
hexan-1-ol+benzyl alcohol[23] 40 1.23 1.65 1.49 1.46 1.65 1.62 1.34
50 2.26 1.54 1.70 1.87 1.53 1.60 1.97
60 7.37 6.38 6.55 6.83 6.26 6.43 7.02
methanol + ethanol[18] 10 1.90 2.05 1.97 2.01 2.30 1.96 1.73
20 1.01 0.90 1.00 1.00 1.13 0.99 0.97
30 0.91 0.64 0.84 0.81 0.90 0.84 0.73
40 1.23 0.98 1.13 1.08 1.39 1.09 0.78
50 1.21 1.11 1.10 1.04 1.71 1.00 0.46
acetone + water[18] 20 1.62 12.77 6.39 1.64 13.20 6.78 2.42
30 1.13 11.15 4.86 1.14 11.59 5.17 1.93
37.8 1.41 10.44 4.46 1.54 10.90 4.83 2.08
40 1.99 9.524 4.00 2.16 10.02 4.38 2.47
ethanol + water[18] 10 5.92 5.50 5.23 6.37 6.04 5.26 7.64
20 4.17 3.64 3.49 4.44 4.12 3.51 5.71
40 5.07 4.07 4.02 5.25 3.80 4.37 6.26
50 6.91 6.17 6.01 7.24 6.04 6.54 8.23
methanol + water[18] 10 3.02 2.90 1.54 3.36 3.16 1.65 2.86
20 3.18 2.80 2.02 3.38 2.95 2.14 3.06
40 2.35 1.76 1.55 2.39 1.74 1.45 2.45
50 3.95 3.47 2.67 4.03 3.03 2.62 3.89
overall relative deviation 2.53/2.70 3.43/2.54 2.70/2.44 2.63/2.74 3.52/2.59 2.76/2.47 2.78/2.84
Overall relative deviation excluding the system acetone + water.

using the binary parameters obtained from the viscos-
ity isotherm of binary solutions. Eighteen ternary sys-
tems including 29 isotherms are investigated with the
binary parameters obtained in the calculation carried
out in section 3.1. The relative deviations of the pre-
dicted values of viscosities from the experimental
ones are summarized in Table 3. The prediction using
the equations of state together with the gE-mixing rule
is successful not only for polar systems but also for
the aqueous solutions. The overall average deviations
of the oC3EOS with either the 3p or the 2p variants of
the gE-mixing rule are less than 2%. It is to be noted
that the isotherms at temperatures above 30 for the
last four systems are extrapolated using the binary pa-
rameters obtained at 30. The results of the extrapola-
tion are also satisfactory. Fig.4 illustrates the viscosities
of the system chlorobenzene+hexan-1-ol+benzyl alco-
hol at different temperatures predicted using the
oC3EOS. The solid and dashed lines represent results
by using the 3p and 2p gE-mixing rule, respectively

Chinese J. Ch. E. 14(6) 770 (2006)

776 Chinese J. Ch. E. (Vol. 14, No.6)

Table 3 Prediction of viscosities for ternary systems

, oC3EOS PREOS PTEOS

System
rel ( )VL rel ( )2p rel ( )3p rel ( )VL rel ( )2p rel ( )3p rel ( )RK

acetone+ethanol+propan-2-ol[34] 25 9.76 4.16 2.47 9.92 4.43 1.35 2.92

acetone+ethanol+methanol[34] 25 7.82 3.02 3.58 8.30 3.01 3.96 2.60
[34]
acetone+hexane+cyclohexane 25 1.72 1.28 1.34 5.35 0.95 3.66 1.39
[34]
acetone+hexane+ethanol 25 1.85 1.18 1.04 3.69 1.03 1.34 1.42
acetone+carbon tetrachloride+cyclohexane[34] 25 1.58 0.53 0.77 3.29 0.60 0.47 0.63
[34]
dimethylsulfoxide+chloroform+methanol 25 7.83 1.79 1.44 3.15 1.75 1.26 5.71
ethanol+acetone+cyclohexane[34] 25 2.13 1.51 1.41 4.53 1.31 2.16 3.04
ethanol +cyclohexane+propan-2-ol[34] 25 4.73 1.47 1.82 6.13 2.07 1.90 1.69
hexane+cyclohexane+ethanol[34] 25 2.41 1.57 1.28 4.09 1.38 4.35 2.16
[18]
methanol+ethanol+propan-1-ol 30 2.32 1.23 0.78 2.24 0.85 0.84 0.88
methanol+ethanol+propan-2-ol[34] 25 3.95 2.00 1.44 2.67 2.01 2.59 1.78
[34]
tetrachloromethane+cyclohexane+propan-2-ol 25 5.23 0.82 0.73 5.84 3.22 1.24 1.86
ethanol+propan-1-ol+water[18] 30 55.24 2.53 2.52 45.95 4.20 1.65 9.47
[18]
methanol+propan-1-ol+water 30 46.98 1.69 1.94 38.57 1.48 2.34 6.88
30 35.30 1.18 1.85 36.51 1.25 1.94 5.24
methanol+ethanol+water[18] 40 36.30 2.50 1.16 36.85 2.92 1.31 4.58
50 37.20 4.92 2.89 37.54 5.72 3.34 4.46
30 7.11 2.19 2.30 8.00 2.22 2.86 2.89
40 7.28 3.17 3.29 8.21 3.22 3.83 3.37
chlorobenzene+hexan-1-ol+benzyl alcohol[35]
50 4.85 1.36 1.48 5.83 1.43 2.02 1.10
60 5.24 2.40 2.51 6.22 2.48 3.03 2.14
30 3.52 1.83 1.55 5.96 1.87 1.69 3.06
40 2.25 0.85 0.85 0.72 0.84 0.78 1.86
hexane+benzyl alcohol+chlorobenzene[25]
50 3.86 1.82 1.67 1.32 1.84 1.65 3.29
60 2.31 1.86 1.88 1.14 1.88 1.75 2.79
30 1.03 1.43 1.36 3.05 1.60 0.92 1.85
40 2.47 1.55 1.58 4.55 1.51 1.75 2.25
hexane+benzyl alcohol+hexan-1-ol[25]
50 1.07 0.93 0.89 2.66 1.02 0.64 1.67
60 2.37 2.34 2.39 3.84 2.29 2.39 2.88
overall average deviation 10.92 1.90 1.71 10.93 2.08 2.01 2.97

and the dotted lines represent the results using the van
Laar-mixing rule. The 3p gE-mixing rule reproduces
the experimental data very well over the whole tem-
perature range. The prediction using the 2p gE-mixing
rule is satisfactory at low temperatures, whereas its de-
viation from experimental data increases with the in-
crease in temperature. For the asymmetric van
Laar-mixing rule, there exist obvious systematic devia-
tions of the calculated values from the experimental
ones. Such deviations become even larger for aqueous
ternary systems. As shown in Fig.5, by the example
system methanol+ethanol+water, the deviations are
very large with an average error of more than 35%,
which means that the van Laar-mixing rule is not suit-
able for such calculations, particularly, for associated
systems. On the contrast, the prediction results of nor-
December, 2006
mal systems obtained by using the RK mixing rule are
similar to those by using gE-mixing rule, however, for
aqueous systems, it gives relatively larger deviations.
4 SUMMARY
The cubic equations of state have been incorpo-
rated into the absolute rate theory of Eyring to calcu-
late the viscosity of liquid mixtures. A modified
Huron-Vidal gE-mixing rule based on the zero pres-
sure reference state, in which the NRTL equation is
used, has been employed in the calculation in com-
parison with the van Laar and the RK mixing rule.
With three adjusted model parameters, the equations
of state with the gE-mixing rule can give a very satis-
factory viscosity correlation for either nonpolar or
polar systems and even for aqueous solutions. In
 Calculation of Viscosities of Liquid Mixtures Using Eyrings Theory in Combination with Cubic Equations of State 777

results are obtained in the extrapolation of viscosities

of ternary systems using only binary parameters ob-
tained at a given temperature. The asymmetric van
Laar and RK mixing rule can be used very success-
fully to correlate and extrapolate the liquid viscosities
of binary systems, although the van Laar-mixing rule
is unsuitable for the viscosity prediction of multi-
component systems, particularly of aqueous systems.
The RK mixing rule can be used to predict the vis-
cosities of ternary solutions reasonably well for
nonaqueous systems, however, for aqueous systems, it
gives relatively poor results. In addition, there is no
considerable difference in the calculation of dynamic
viscosities among the three cubic EOS used, although
Figure 4 Prediction of viscosities of the system their description of the pvT-behavior is quite different.
chlorobenzene+hexan-1-ol+benzyl alcohol from The density of liquids and liquid mixtures is not sig-
information of binary systems at 25 using the nificant in the calculation of dynamic viscosities.
oC3EOS with the 3p and 2p variants of the
gE-mixing rule and the van Laar-mixing rule
x1=0.068, x2=0.552; x1=0.154, x2=0.413;
NOMENCLATURE
x1=0.272, x2=0.492; x1=0.347, x2=0.050;
Aij energy parameters of the NRTL model in the
x1=0.413, x2=0.118; x1=0.497, x2=0.353; gE-mixing rule
x1=0.569, x2=0.224; x1=0.641, x2=0.296; a, b, u parameters in the cubic 3-parameter equation of
E E
oC3EOS 3p g ; - - - - oC3EOS 2p g ; ------- oC3EOS VL state oC3EOS
Fobj objective function
gE molar excess Gibbs free energy, Jmol1
E
g dimensionless excess Gibbs free energy
E E
[ g = g /( R T ) ]
g# molar activation Gibbs energy of flow, Jmol1
g#E molar excess activation Gibbs energy of flow, Jmol1
h Planck constant
hE molar excess enthalpy, Jmol1
k reciprocal of the packing fraction
M molar mass
m1, m2, m3 parameters in the (Tr)-function
N number of data points
NA Avogadro constant
p pressure, kPa
R molar gas constant (8.314Jmol1K1)
T thermodynamic temperature, K
v molar volume, dm3mol1
Figure 5 Prediction of viscosities of the system xi mole fraction of component i in liquid mixtures
methanol+ethanol+water from information of binary z compressibility factor
systems at 25 using the oC3EOS with the 3p and 2p zc,opt apparent (optimized) critical compressibility factor
variants of the gE-mixing rule and the van
ij nonrandomness parameter of NRTL model
Laar-mixing rule
x1=0.070, x2=0.241; x1=0.209, x2=0.511; (Tr) temperature function of CEOS parameter a
x1=0.153, x2=0.263; x1=0.258, x2=0.297; average deviation
x1=0.552, x2=0.190; x1=0.554, x2=0.054; dynamic viscosity
x1=0.710, x2=0.061; x1=0.850, x2=0.033; temperature,
E E
oC3EOS 3p g ; - - - - oC3EOS 2p g ; ------- oC3EOS VL proportional factor

i fugacity coefficient of component i in a mixture
majority of the cases the nonrandomness parameter oi
*
fugacity coefficients of pure component i
in the gE-mixing rule can be set as a constant of 0.2
a reduced cohesive energy parameter
without loss of accuracy. This EOS method can be b reduced volumetric parameter
successfully used to predict viscosities of binary sys- Superscripts
tems over a certain temperature range using the pa- E excess property
rameters obtained from a viscosity isotherm at a given id ideal solution
low temperature. It can also be successfully used to Subscripts
predict the liquid viscosities of ternary systems using c critical
binary parameters even for aqueous systems. Good cal calculated value

Chinese J. Ch. E. 14(6) 770 (2006)

778 Chinese J. Ch. E. (Vol. 14, No.6)
exp experimental value
i component i
j component j
k component k
l component l
2p gE-mixing rule with constant nonrandomness
3p gE-mixing rule with constant nonrandomness
RK Redlich-Kister-mixing rule
r reduced value
rel relative deviation
VL van laar-mixing rule
REFERENCES
1 Grunberg, L., Nissan, A.H., Mixture law for viscosity,
Nature, 164, 799802(1949).
2 Glasstone, S., Laidler, K.J., Eyring, H., The Theory of
Rate Processes, McGraw-Hill, New York (1941).
3 Poling, B.E., Prausnitz, J.M., OConnell, J.P., The Proper-
ties of Gases and Liquids, 5th ed., McGraw-Hill, Boston
(2001).
4 Teja, A. S., Rice, P., Generalized corresponding states
method for the viscosities of liquid mixtures, Ind. Eng.
Chem. Fundam., 20, 7781(1981).
5 Liu, H., Wang, W., Chang, C.H., Model with tempera-
ture-independent parameters for the viscosities of liquid
mixtures, Ind. Eng. Chem. Res., 30, 16171624(1991).
6 Papaloannou, D., Evangelou, T., Panayiotou, C., Dy-
namic viscosity of multicomponent liquid mixtures, J.
Chem. Eng. Data, 36, 4346(1991).
7 McAllister, R.A., The viscosities of liquid mixtures,
AIChE J., 6, 427431(1960).
8 Wei, I. C., Rowley, R. L., A local composition model for
multicomponent liquid mixture shear viscosity, Chem.
Eng. Sci., 40, 401408(1985).
9 Wu, D. T., Prediction of viscosities of liquid mixtures by
a group contribution method, Fluid Phase Equilibria,
30, 149156(1986).
10 Chevalier, J. L. E., Petrino, P. J., Gaston-Bonhomme, Y.,
Estimation method for the kinematic viscosity of a liq-
uid-phase mixture, Chem. Eng. Sci., 43, 1303
1309(1988).
11 Cao, W., Fredenslund, A., Rasmussen, P., Statistical ther-
modynamic model for viscosity of pure liquids and liquid
mixtures, Ind. Eng. Chem. Res., 31, 26032619(1992).
12 Cao, W., Knudaen, K., Fredenslund, A., Rasmussen, P.,
Simultaneous correlation of viscosity and vapor-liquid
equilibrium data, Ind. Eng. Chem. Res., 32, 2077
2087(1993).
13 Lee, M.J., Chiu, J.Y., Hwang, S.M., Lin, H.M, Viscosity
calculations with the Eyring-Patel-Teja model for liquid
mixtures, Ind. Eng. Chem. Res., 38, 28672876(1999).
14 Adachi, Y., Sugie, H., A new mixing rulemodified
conventional mixing rule, Fluid Phase Equilibria, 28,
103108(1986).
15 Michelsen, M.L., Kistenmacher, H., On composi-
tion-dependent interaction coefficients, Fluid Phase
Equilibria, 58, 229230(1990).
16 Stryjek, R., Vera, J.H., PRSV an improved Peng-Robinson
equation of state with new mixing rules for strongly non-
ideal mixtures, Can. J. Chem. Eng., 64, 334340(1986).
17 Wei, I.C., Rowley, R.L., Binary liquid viscosities and
densities, J. Chem. Eng. Data, 29, 332335(1984).
18 Dizechl, M., Marschall, E., Viscosity of some binary
and ternary liquid mixtures, J. Chem. Eng. Data, 27,
358363(1982).
19 Chevalier, J. L. E., Pertrino, P. J., Gaston-Bonhomme, Y.
H., Viscosity and density of some aliphatic, cyclic, and
December, 2006
aromatic hydrocarbons binary liquid mixtures, J. Chem.
Eng. Data, 35, 206212(1990).
20 Kouris, S., Panajiotou, C., Dynamic viscosity of mix-
tures of benzene, ethanol, and n-heptane at 298.15K, J.
Chem. Eng. Data, 34, 200203(1989).
21 Asfour, A.F.A., Siddique, M.H.J., Vavanellos, T.D.,
Kinematic viscosity-composition data for eight binary
systems containing toluene or ethylbenzene and C8-C16
n-alkanes at 293.15 and 298.15K, Chem. Eng. Data, 35,
199201(1990).
22 Fermeglia, M., Lapasin, R., Excess volumes and vis-
cosities of binary mixtures of organics, J. Chem. Eng.
Data, 33, 415417(1988).
23 Singh, R.P., Sinha, C.P., Viscosity and thermodynamics
of viscous flow of some associated mixtures, Z. Phys.
Chemie, Leipzig, 265, 593602(1984).
24 Asfour, A.F.A., Dullien, F.A.L., Viscosities and densi-
ties of four binary liquid systems at 25.00, J. Chem.
Eng. Data, 26, 312316(1981).
25 Singh, R.P., Sinha, C.P., Singh, B.N., Viscosities densi-
ties and activation energies of viscous flow of some ter-
nary systems and their partially miscible binary subsys-
tem n-hexane+benzyl alcohol at 30, 40, 50, and 60, J.
Chem. Eng. Data, 31, 107111(1986).
26 Orge, B., Iglesias, M., Rodrigguez, A., Canosa, J.M., Tojo,
J., Mixing properties of (methanol, ethanol, or
1-propanol) with (n-pentane, n-hexane, n-heptane and
n-octane), Fluid Phase Equilibria, 133, 213
227(1997).
27 Aminabhavi, V.A., Aminabhavi, T.M., Balundgi, R.H.,
Evaluation of excess parameters from densities and
viscosities of binary mixtures of ethanol with anisole,
N,N-dimethylformamide, carbon tetrachloride, and ace-
tophenone from 298.15 to 313,14K, J. Chem. Eng. Data,
29, 21062111(1990).
28 Singh, R.P., Sinha, C.P., Viscosities and activation en-
ergies of viscous flow of the binary mixtures of n-hexane
with toluene, chlorobenzene, and 1-hexanol, J. Chem.
Eng. Data, 29, 132135(1984).
29 Hammond, L.W., Howard, K.S., McAllister, R.A., Vis-
cosities and densities of methanol-toluene solutions up to
their normal boiling points, J. Phys. Chem., 62, 637
639(1958).
30 Katti, P.K., Chaudhri, M.M., Prakash, O., Viscosities of
binary mixtures of carbon tetrachloride with methanol
and isopropyl alcohol, J. Chem. Eng. Data, 11, 593
594(1966).
31 Christensen, C., Gmehling, J., Rasmussen, P., Weidlich,
U., Heat of Mixing Data Collection, Binary Systems,
DECHEMA Chemistry Data Series, Vol. III, Part 1,
Part2, Frankfurt (1984).
32 Ji, W.R., Lempe, D.A., A systematic study of cubic
three-parameter equations for deriving a structurally op-
timized PVT relation, Fluid Phase Equilibria, 147,
85103(1998).
33 Ji, W.R., Stiebing, E., Lempe, D.A., Strukturooptimierte
Dreiparameter-Zustandsgleichung (C3EOS) zur Besch-
reibung von Phasengleichgewichts und anderen thermo-
physikalischen Eigenschaften. Bericht zum DFG-
Vorhaben LE 886/4-1, Merseburg (2004).
34 Wei, I.C., Rowley, R.L., Ternary liquid mixture viscosi-
ties and densities, J. Chem. Eng. Data, 29, 336
340(1984).
35 Singh, R.P., Sinha, C.P., Viscosities and activation ener-
gies of viscous flow of the ternary mixtures of toluene,
chlorobenzene, and 1-hexanol, and benyl alcohol, J.
Chem. Eng. Data, 30, 470474(1985).
 Calculation of Viscosities of Liquid Mixtures Using Eyrings Theory in Combination with Cubic Equations of State 779

APPENDIX a c v + c c d
1 Cubic 3-parameter equation of state oC3EOS[32] n ln
RT (c d ) ni 2
v + d v + c
RT a (T )
p= (A1)
v b v 2 + ub ( v b ) d v + c c d
n ln (A10)
RT ni v + d v + d
R 2Tc2
a (T ) = ac (Tr ) = a (Tr ) b = bc = b c (A2) 2 Modified Huron-Vidal gE-mixing rule[33] based on zero
pc pc
pressure reference state
where the dimensionless cohesive energy and volumetric pa-
cd
rameter and the parameter u are estimated using the apparent a (T ) =
kb + c
(optimized) compressibility factor zc, opt: ln
kb + d
3 2 3
b + (1 3 zc,opt ) b + 3 zc,opt b zc,opt = 0 (A3) kb + ckk


a (T ) b
xk kk ln kk RT xk ln RTg 0E
a = 1 3 zc,opt (1 zc,opt ) + 3(1 2 zc,opt ) b + 2 b2 (A4) ckk d kk kbkk + d kk bkk
k k
u = (1 + b 3 zc,opt ) / b (A5) (A11)
where the packing-fraction (1/k) is set as (1/1.1) and the excess
A three-parameter (Tr)-function is used in this study

{
Gibbs free energy at zero pressure is estimated using the NRTL
(Tr ) = 1 + m1 (1 T 0.75 ) + equation:
r

x G
(A6)
)}


( ) ( ) (
0.75 0.7 T 0.75
2 l lk lk
m 0.75 + m
gE
2 1 Tr 3 1 Tr
x G
r
g 0E = 0 = xk l (A12)
The zc,opt and the parameters in the (Tr)-function (m1, m2 and RT l lk

k
m3) are fitted using experimental saturated liquid densities and l
vapor pressures of pure substances. In the calculation, a generic Aij
form of cubic equation of state is used
ij = Gij = exp( ij ij ) ( ij = ji ) (A13)
T
RT a (T ) For parameter b and u, a linear combination is used.
p= (A7)
v b (v + c)(v + d ) 3 van Laar-type mixing rule
where the parameter c and d for the oC3EOS are estimated by
x x (a
kkl klk
u u (u + 4) u + u (u + 4) a= k l kk all )1/ 2 1 (A14)
c = b d = b xk kkl + xl klk
(A8) k l
2 2
4 Patel-Teja equation of state
and for PREOS by
RT a (T )
c = (1 2 ) b d = (1 + 2 ) b (A9) p= (A15)
v b v (v + b ) + c (v b )
The fugacity coefficient can be calculated by
5 Redlich-Kister-type mixing rule
p (v b ) 1 b
ln i* = ln
RT + z 1+ n
v b ni


a= x x (a
k l
k l kk all )1/ 2 1 kkl klk ( xk xl ) (A16)

1 a v + c
a + n ln +
RT (c d ) ni v + d

Table A1 Parameters of binary systems used for the prediction of viscosities of

ternary aqueous systems using the oC3EOS
Van Laar-type mixing rule 2p gE-mixing rule 3p gE-mixing rule RK mixing rule
System
kkl klk A12/K A21/K 12 A12/K A21/K kkl klk
ethanol+propan-1-ol 0.028 0.027 414.57 385.01 2.088 346.10 140.53 0.083 0.080
methanol+ethanol 0.065 0.045 118.82 73.39 0.143 151.09 393.84 0.025 0.041
methanol+propan-1-ol 0.139 0.128 184.15 585.89 0.262 1.20 442.36 0.083 0.080
ethanol+water 1.384 0.661 2675.04 1167.64 0.188 2773.53 1234.21 1.245 0.541
methanol+water 0.823 0.523 1953.22 168.19 0.127 2952.48 3035.73 0.824 0.546
propan-1-ol+water 1.555 0.773 2810.51 1478.51 0.255 2511.57 952.569 1.252 0.435
12 set to be 0.2.
Parameters for PT EOS.

Chinese J. Ch. E. 14(6) 770 (2006)